[Box Backup] Mailbox backup is dangerous.

[scartomail]

--- On Fri, 8/28/09, Chris Wilson <chris@qwirx.com> wrote:

> From: Chris Wilson <chris@qwirx.com>
> Subject: Re: [Box Backup] Mailbox backup is dangerous.
> To: boxbackup@boxbackup.org
> Date: Friday, August 28, 2009, 6:33 PM
> Hi Edo,
>=20
> On Fri, 28 Aug 2009, scartomail wrote:
>=20
> > @Chris(or other developpers): Would you please give
> you opinion on this?
>=20
> I'm not sure that it is a "bug". Box Backup was originally
> designed to work this way. The procedure for removing old
> and deleted versions is undocumented and I don't claim to
> understand it. A bug by my definition is when software's
> behaviour does not match its documentation, or it fails to
> perform a documented function.

Very true, and scary.
We have software running around and we do not know what it does.

>=20
> I agree that the behaviour is undesirable. I have been
> intending for a long time to implement backup snapshots
> where every file within a snapshot is preserved, and the
> client, not the server, chooses when to delete an entire
> snapshot to free up space on the store.


BB does have 2 bakup mode's snapshot and lazy.
Would it be best to use the snapshot for now or will houskeeping kick in an=
yway?

>=20
> Some members have written patches to change the
> housekeeping deletion behaviour. However as I don't
> understand it, I don't feel qualified to review those
> patches, and I feel that the current housekeeping behaviour
> is fundamentally undesirable, and patching it will not help.
> The only thing that would make me (personally) comfortable
> that my old versions of files were being preserved to my
> liking is the above immutable snapshot behaviour.
>=20

Ok, this is me thinking out load with no real c++ programming skils.
BB does 3 things.
- backup: this seems to work
- housekeeping : we don't realy know what it does and it sometimes deletes =
files.
- restore: this seems to work

Let's say housekeeping is currently doing 10 things.
Why don't we throw away the housekeeping part and replace it with a new hou=
skeeping that just does 1 or 2 things that we know of it does well and are =
essential for BB to function.
We create that by copy and paste the housekeeping stuff we know and underst=
and.

This way we have something we understand, is stabel and we can always add f=
unctionality to it as we go allong.

If this is a ridiculous way of looking at the problem, I'm sorry.
Remember I was just thinking out loud.


> > If there are bugs in BB and it isn't stable yet,
> that's ok, it is still under development. But a lot of
> people report all kinds of bugs and errors with BB.
>=20
> Most of the bugs reported are not data loss bugs, e.g.
> failure to compile, bbackupd crashes, or new feature
> requests.
>=20
> I try to investigate data loss bugs as best I can, if I
> don't understand why they occur or if there's a chance of
> retrieving the lost data. Unfortunately in Tom's case I have
> already failed on many attempts to understand the
> housekeeping deletion logic, and there is no chance of
> recovery of the lost data.
>=20
> I also haven't had a lot of time to attend to Box over the
> last two years. When I joined the project, I originally
> intended just to write Boxi. Since then I have stepped up to
> maintaining the windows port, and then to maintaining the
> whole of Box Backup.
>=20
> > Shouldn't we test and address(report) them(or are they
> already reported)?
>=20
> Yes, definitely, please do report them, test them, and hold
> us to account to make sure that we publish correct,
> informative and not misleading information about the
> capabilities and omissions or risks of Box Backup on our
> website and other appropriate places.

Will do.
I'm currently setting up the test environment as I said before.
I will try to run several scenario's to see whappens.

Most problmems of data loss seem to be over a long period of time.
I'll get back to you on doing the same but than setting it up(configuring) =
for a day or a week tops.

Anyway thanks for the clear anwser, it's much apreciated.

Rgds Edo


>=20
> Cheers, Chris.
> -- _____ __=A0 =A0=A0=A0_
> \=A0 __/ / ,__(_)_=A0 | Chris Wilson <0000 at
> qwirx.com> - Cambs UK |
> / (_/ ,\/ _/ /_ \ | Security/C/C++/Java/Ruby/Perl/SQL
> Developer |
> \__/_/_/_//_/___/ | We are GNU : free your mind & your
> software |
> _______________________________________________
> boxbackup mailing list
> boxbackup@boxbackup.org
> http://lists.warhead.org.uk/mailman/listinfo/boxbackup
> =0A=0A=0A